Testing Medusa: What to Test, What to Skip, and How
Commerce code moves money, so untested checkout logic is a liability. A pragmatic test strategy — what earns its keep, and what is theatre.
Articles tagged DevOps — hands-on notes from the commerce builds we ship.
Commerce code moves money, so untested checkout logic is a liability. A pragmatic test strategy — what earns its keep, and what is theatre.
Self-hosting means owning the question 'is checkout working?'. Structured logging, tracing, the four metrics that matter and alerts that do not cry wolf.
Four cache layers, one rule about carts, and the invalidation strategy that stops a price change taking twelve hours to appear.
Traffic does not break Medusa — the database does, then the workers, then the catalog queries. The bottlenecks in the order you will meet them.
A production AWS architecture for Medusa — ECS Fargate, RDS, ElastiCache, S3 — plus the networking and migration details that turn a two-day job into a two-week one.
Railway is the shortest path from a Medusa repository to a production deployment that is actually correct. The full setup, including the worker service everyone forgets.
A multi-stage Dockerfile for Medusa v2, a compose file for local development, and the build-memory problem that catches every team once.
Every variable Medusa reads, which ones are security-critical, and how to structure configuration so a missing value fails at boot instead of at checkout.
Generated migrations are only as safe as your review of them. How Medusa migrations work, and the expand-and-contract pattern that keeps rolling deploys from failing.
What a production Medusa deployment actually needs — process separation, Redis, migrations, health checks — and the four mistakes that cause the first outage.